Procedia Engineering 48 ( 2012 ) 352 – 357
1877-7058 © 2012 Published by Elsevier Ltd.Selection and/or peer-review under responsibility of the Branch Offi ce of Slovak Metallurgical Society at Faculty of Metallurgy and Faculty of Mechanical Engineering, Technical University of Košice
doi: 10.1016/j.proeng.2012.09.525
MMaMS 2012
Control of a test bench for dynamic emulation of mechanical loads
Karol Kyslan
a*
,František Ćurovský
aa
Department of Electrical Engineering and Mechatronics, Technical University of Košice, Letná 9, 042 00 Košice, Slovak Republic
Abstract
The paper describes a control strategy for a test bench consisting of a drive under test mechanically connected with a dynamometer drive. The control of dynamometer drive is performed on RT-Lab simulation platform and based on dynamic emulation of mechanical loads (DEML) strategy. Torque-controlled dynamometer emulates the behaviour of a real mechanical load. The test bench enables emulating of the linear and non-linear dynamics of mechanical loads with only mathematical model of the load included into the closed-loop control. Main purpose of the research on this kind of test bench’s control is a simplification of manufacturer tests on new actuator prototypes. The next equally important purpose is reducing the commissioning time of drives’ control. Simulation and experimental results showing the abilities of the proposed test bench‘s control approach are presented.
© 2012 The Authors. Published by Elsevier Ltd.
Selection and/or peer-review under responsibility of the Branch Office of Slovak Metallurgical Society at Faculty of Metallurgy and Faculty of Mechanical Engineering, Technical University of Košice.
Keywords: dynamic emulation; mechanical load; dynamometer; RT-Lab; industrial converter
Nomenclature
Bem emulated viscous friction coefficient (Nms/rad) Tfr Coulomb friction torque (Nm)
cĭ torque constant (Nm/A) TLref reference torque for load drive (Nm)
g gravity acceleration (m/s2) T
load actual torque of load drive (Nm) J total real shaft inertia incl. both machines (kgm2) TRI time constant of current controller (s) Jem emulated inertia (kgm
2
) TTM converter time constant (s)
Ka rotor gain (ȍ
-1
) Ttest actual torque of drive under test (Nm)
Ktm converter gain (V/V) Ttar actual torque of target drive (Nm)
KRI current controller gain Ȧ rotor angular speed (rad/s)
lem emulated length (m) Ȧem emulated angular speed (rad/s)
mem emulated mass (kg) Ȧref reference angular speed from ramp generator (rad/s)
Tcom compensating torque (Nm) Ia separately excited machine armature current (A)
Te drive under test’s speed controller output (Nm) ijem emulated rotor angle or position (rad) Te tar target drive’s speed controller output (Nm) Ta rotor time constant (s)
Text external torque applied to motor or load (Nm) Va rotor voltage (V) Vemf back emf voltage (V)
* Corresponding author. Tel.: +421 55 602 2155; E-mail address: [email protected]
© 2012 Published by Elsevier Ltd.Selection and/or peer-review under responsibility of the Branch Offi ce of Slovak Metallurgical Society at Faculty of Metallurgy and Faculty of Mechanical Engineering, Technical University of KošiceOpen access under CC BY-NC-ND license.
1. Introduction
Emulation is an imitation of the system or its part by another platform or technical device in such a way that the imitating system behaves similarly like the imitated one. If the input data of the imitated and imitating system are the same, also the output data of both systems should have exactly the same values. The imitating system is called emulator and imitating technique is called emulation. Dynamic emulation of mechanical loads (DEML) is a common name for advanced dynamometer strategies which has been developed in last few years. Authors in [1], [2] or [3] use a different approaches for the dynamometer control. The aim is to control the dynamometer in such way that it behaves like real mechanical load connected to the drive under test.
In the later phase of a development and testing of control algorithms for various mechanical loads there is a need for detailed experimental validating of control algorithm. When possible, real load is connected and algorithm is tested. But in the case of high power loads or complex mechanical loads (production lines etc.) they are not available in the laboratory conditions. Nevertheless, there is demand for testing of algorithm even if mechanical load is not available. The solution is to connect drive under test to the dynamometer emulating behaviour of mechanical load. Thus, dynamometer should have advanced torque controller with the model of emulated load included in the control structure. Machine under test mechanically connected with dynamometer on the rigid shaft with the converters and control device creates the test bench, simply termed as emulator.
The paper describes a simulation model of a test bench in Section II and its implementation into the closed-loop HIL structure in Section III. Simulation and experimental results are covered in Section IV. Finally, conclusions and proposals for the future work are presented in Section V.
2. Simulation model of proposed emulator
Simulation model of emulator created in Matlab/Simulink environment is in the upper part of Fig. 1 (green). Furthermore, there is a model of so called target system in the lower part of Fig.1 (blue). Target system is the drive with connected real mechanical load. The same reference value from ramp generator is entering both emulator and target drive. Emulated speed Ȧem on the target drive’s shaft is compared with real speed Ȧ on the emulator’s shaft. If the speed controllers are the same and implemented DEML algorithm is working precisely, both speeds have to be the same in the steady state and transients.
The blocks speed controllers are simple PI controllers and I-components of both PI controllers are limited. The outputs of the both speed controllers are assumed to be torque references. In the case of target system, limited torque reference is entering the block torque loop. This block contains all the dynamics from limited reference torque to actual shaft torque, i.e. current controller, converter and machine dynamics as in Fig. 2(a). From Fig. 2(b) it is clear, that non-linear emulator shaft is assumed. Coulomb friction component Tfr is acting on the shaft and therefore it has to be compensated by emulator control.
The reference torque TLref for dynamometer (load drive) is calculated in the block DEML algorithm+model of mechanical
load. Its value is derived from the model of mechanical load. Further information about DEML algorithm can be found
in [4]. Compensating torque Tcom acts against friction torque in the emulator’s mechanics. After compensation of Tfr by Tcom linear mechanics of emulator can be assumed.
Fig. 1. Simulink model of a test bench with target system for comparison of the speed and torque responses
DEML ALGORITHM + MODEL OF MECHANICAL LOAD TEST BENCH MECHANICS T load Ttest w TARGET SYSTEM TORQUE LOOP Te tar w act T tar TARGET SYSTEM SPEED CONTROLLER w ref w em Te tar RAMP GENERATOR T ext w ref MECHANICAL LOAD T tar T ext w_em LOAD DRIVE TORQUE LOOP 1 0.002s+1 T ext w act Te w em T Lref T com
DRVE UNDER TEST TORQUE LOOP
Te
w act T test
DRIVE UNDER TEST SPEED CONTROLLER
w em
w ref Te
Fig. 2. Model of a target system’s torque loop (a) emulator mechanics' model (b) The model in Fig.1 uses these main simplifications:
• it is assumed that high dynamic vector-controlled drive is used for dynamometer and therefore dynamics of whole load drive is replaced only by first order lag model,
• there is no backlash effect in the emulator’s shaft,
• there is no transport delay caused by sampling time of used converters
Separately excited DC machine is used as a drive under test. Obviously parameters of the target system machine are the same as parameters of drive under test machine. Therefore, torque loops of both machine under test and target machine are the same. It is assumed that DC machine does not work in field-weakening region and therefore it is assumed as a machine with constant flux. In that way the model in Fig. 2a is correct.
Block mechanical load contains the dynamics of mechanical load, which parameters can be set by the user. In this paper the results are shown for simple linear load and for simple robotic arm as highly nonlinear load. The dynamics of simple linear load can be described as:
em tar ext em em em
d
T
T
B
dt
J
ω
=
−
−
ω
(1)where user parameters are Jem, Bem and Text. Dynamics of a simple robotic arm can be described as:
sin(
)
;
em tar ext em em em em em em em emd
T
T
B
m gl
d
dt
J
dt
ω
=
−
−
ω
−
ϕ
ϕ
=
ω
(2)where user parameter are Jem, Bem, Text, mem, and lem. Both linear load and robotic arm models are in Fig. 3(a) and Fig. 3(b).
Fig. 3. Dynamics of (a) simple linear load (b) robotic arm
converter dynamics machine dynamics Ia Va V emf PI current controler I ref T tar 1 c.phi
1/c.phi KrI cfi
Ktm Ttm.s+1 1 TiI.s Ka Ta.s+1 w act 2 Te tar 1 w 1 T fr 1 J.s 0 |u| T test 2 T load 1 (a) (b) phi_em 2 w_em 1 sin 1 s 1 s m_em*g*l_em 1/((J_em+m_em*l_em^2)) B_em T ext 2 T tar 1 w_em 1 1 s 1/J_em B_em T ext 2 T tar 1 (a) (b)
3. Implementation of emulator into the HIL structure
HIL structure with emulator consists of two machines on common shaft and connected with clutch. Separately excited DC machine acts as drive under test and it is supplied by Siemens Simoreg 6RA70 converter. Induction machine acts as a load drive – dynamometer and it is supplied by Siemens Simovert Master Drives 6SE70 converter. The communication between both converters is based on Simolink protocol with optical-fibre cable as a transfer medium. This protocol enables highly dynamic and accurate synchronism of both converters [6].
As the higher-level control the distributed real-time platform RT-Lab in configuration Target PC – Command Station was used. Control algorithm was developed in Matlab/Simulink environment and experiments were realized with sampling time of 250ȝs. Communication between Target PC and Simoreg converter is based on CAN protocol and CAN-ACx-PCI data acquisition card was used on Target PC side. Connection of industrial converter and Matlab environment running on Target PC enables a variety of experiments not only for this application. Realization of the test bench in detail is not presented here and only experimental results are included.
4. Simulation and experimental results
Simulation results of simple linear load emulation are presented in Fig. 4 and Fig. 5. When comparing torque responses in Fig. 4 and Fig. 5, it can be observed that even if there is no external torque Text applied, values of accelerating and decelerating torque in Fig. 5 are higher than that in Fig 4. It is caused because of inertia value was increased from 3J in Fig. 4 to 8J in Fig. 5. It should be noted that parameters of speed controllers remain unchanged in both cases. Considering the speed error as difference between actual speed on emulator’s shaft and emulated speed of target drive, speed error stays within the band of ±0,5%.
Simulation results of robotic arm emulation are presented in Fig. 6 and Fig. 7. It should be noted that the simple PI controller is used and therefore the results in Fig. 6 and Fig. 7 do not show a good speed reference Ȧref tracking. In order to improve speed reference tracking the structure of the speed controller should be changed. But when considering emulated model tracking Ȧem, the speed errors stays within the band of ±0,5%. Moreover it should be noted that scale for the emulated position is in radians. Responses in Fig. 6 and Fig. 7 shows the state, when robotic arm gets into the position of ijem =45,3 rad. When comparing torque responses, it can be observed that accelerating and decelerating torque value is higher in Fig. 7 than in Fig. 6. It is because of robotic arm’s length lem was increased from 0,3m in Fig. 6 to 0,7m in Fig. 7 with other parameters unchanged.
Experimental results of robotic arm emulation are presented in Fig. 8. It can be observed, that speed error was increased, but it stays within the band of ±3%. Moreover, differences in torque responses can be observed. It is caused by comparing the responses from real emulator device only against the model of target system. The final validation of emulation quality has to be done against real robotic arm responses.
Fig. 4. Linear load emulation, Jem=3J, Text=0 Nm, Bem=0 Nms/rad Fig. 5. Linear load emulation, Jem=8J, Mext=0, Bem=0 Nms/rad
0 1 2 3 4 5 6 7 8 9 0 20 40 60 80 sp e e d s [ % ] wref w wem 0 1 2 3 4 5 6 7 8 9 -150 -100 -50 0 50 100 150 time [s] to rq ues [% ] Te Tetar 0 1 2 3 4 5 6 7 8 9 0 20 40 60 80 s p ee ds [%] wref w wem 0 1 2 3 4 5 6 7 8 9 -50 0 50 time [s] to rq u e s [%] Te Tetar
Fig. 6. Simple robotic arm emulation, lem=0,3m, mem=3kg, Bem=0 Nms/rad Fig. 7. Simple robotic arm emulation, lem=0,7m, mem=3kg, Bem=0 Nms/rad
Fig. 8. Experimental results of simple robotic arm emulation, lem=0,3m, mem=3kg, Bem=0 Nms/rad
5. Conclusion
Control strategy of a test bench enabling dynamic emulation of mechanical loads is presented in this paper. Simulation results of linear load and simple robotic arm are included. Realized simulations show a good emulation quality and demonstrate the possibilities of presented emulator. Paper is focused on simulation model of emulator, but included experimental results show the successful HIL implementation of emulator. The differences between simulation and experimental results are caused by some non-linear effects omitted in the emulator’s model.
Future research should include emulation of other types of mechanical loads, extension from speed control loop to position control loop approach and further increasing of emulation quality, which can be done by inserting additional non-linear effects into the model of emulator.
Acknowledgement
This work is the result of the project implementation: Výskum modulov pre inteligentné robotické systémy, ITMS: 26220220141, supported by the Research & Development
0 2 4 6 8 10 12 14 16 0 5 10 speed s [%] w ref w act w em phi em 0 2 4 6 8 10 12 14 16 -40 -20 0 20 40 time [s] to rques [%] Te Tetar ijem=75,36 rad 0 1 2 3 4 5 6 7 8 9 -5 0 5 10 15 s p e eds [ % ] wref w wem phiem 0 1 2 3 4 5 6 7 8 9 -50 0 50 100 time [s] to rq u e s [%] Te Tetar 0 1 2 3 4 5 6 7 8 9 -5 0 5 10 15 sp e e d s [ % ] wref w wem phiem 0 1 2 3 4 5 6 7 8 9 -60 -40 -20 0 20 40 60 80 time [s] to rq u e s [%] Te Tetar
Operational Programme funded by the ERDF. The work was also supported by the Slovak Research and Development Agency under the contract no. APVV-0185-10.
References
[1] Arellano-Padilla, J., Asher, G.M., Sumner, M., 2006. Control of an AC Dynamometer for Dynamic Emulation of Mechanical Loads With Stiff and Flexible Shafts. Industrial Electronics, IEEE Transactions on, June 2006, vol.53, no.4, pp.1250-1260.
[2] Rodic, M., Jezernik, K., Trlep, M., 2006. Mechatronic Systems’ Control Design Using Dynamic emulation of Mechanical Loads. Automatika: Journal for Control, Measurement, Electronics, Computing and Communications, May 2006, vol. 47, no 1-2, pp. 11-18, ISSN 0005–1144.
[3] Žalman, M., Macko, R., 2005. Design and realization of programmable emulator of mechanical loads. 16th IFAC World Congress, July 2005, Volume 16, Czech Republic.
[4] Kyslan, K., Ćurovský, F.. Dynamic Emulation of Mechanical Loads – An Approach Based on Industrial Drives’ Features. unpublished [5] compendium Siemens Simovert Masterdrives 6SE70, available 5/2012 on:
http://support.automation.siemens.com/WW/llisapi.dll?func=cslib.csinfo&lang=en&siteid=cseus&aktprim=0&extranet=standard&viewreg=WW&obji d=10804947&treeLang=en
